Key explanatory variables: Big Five personality traits

Key explanatory variables: Big Five personality traits

Table 2 shows the definitions and descriptive statistics of the variables for the Big Five personality traits. As usual in large surveys (Rammstedt and John 2005, 2007; Soto and John 2017), personality is measured in the pairfam using a short whiplr ekşi version of the Big Five Inventory. Extraversion, conscientiousness, agreeableness, and neuroticism are assessed with four items. Openness to experience is assessed with five items. The items are measured on a 5-point Likert scale ranging from 1 �strongly disagree� to 5 �strongly agree.� For each of the five personality traits, we add up the respective items and divide the sum by the number of items.

Psychologists often assume that the Big Five change only modestly after they have developed in childhood and adolescence (Costa et al. 2000). Some even emphasize an important genetic component contributing to a relatively high stability of personality (Bouchard and Loehlin 2001; Kandler et al. 2010). The question of the stability of the Big Five has also been of interest to economists. Cobb-Clark and Schurer (2012) demonstrate for working-age adults in Australia that mean level changes of the Big Five personality traits are very small over a 4-year period. Considering an 8-year time frame, Elkins et al. (2017) find somewhat higher, but generally small changes even for adolescents and young adults. Anger et al. (2017) confirm for working individuals in Germany that the mean level changes of the Big Five are small over an 8-year period.

As the pairfam provides information on the Big Five for the years 2009, 2013, and 2017, we can examine the stability of the Big Five also with our data. Appendix Table 11 shows that the mean level changes of the Big Five personality traits for the periods , , and are small. The table also reports Cohen’s d. This measure defines the mean level change in terms of a standard deviation change of the respective trait. All values for Cohen’s d amount to less than 0.2 implying that the changes can be considered very small.

In order to account for a nonlinear influence of age on sexuality, we also include a quadratic and a cubic age variable

Altogether, a series of studies suggest that while not literally fixed, the Big Five personality traits exhibit relatively high stability. However, there is an ongoing debate as to the role of age and major life events in intra-individual changes of personality. Specht et al. (2011) find that age has an influence throughout the life span and provide some evidence of a curvilinear influence. The evidence on the role of major life events appears to be mixed. While some studies indicate some moderate and rather specific influences of single life events (Angeli et al. 2018; Anger et al. 2017), others conclude that intra-individual changes are generally only weakly or even not related to major life events and that changes are not economically meaningful (Costa et al. 2000; Cobb-Clark and Schurer 2012). Moreover, Specht et al. (2011) suggest that there can also be reverse causation with personality influencing life events.

Whatever the exact role of life events may be, our dataset allows us to include a rich set of control variables capturing demographic, economic, and family-related factors. This should mitigate endogeneity concerns. We follow most of the studies on personality (e.g., Caliendo et al. 2014; Cobb-Clark and Tan 2011; Mueller and Plug 2006; Risse et al. 2018) and consider personality traits with suitable caution as exogenous. While the regressions may not allow definite causal inferences to be drawn, they provide a crucial first step to bring important new insights to family economics which can be interpreted in light of our theoretical considerations.

Control variables

Appendix Table 12 provides the definitions and descriptive statistics of the control variables. We control for the economic situation by including variables for the years of schooling and the person’s labor market status. A variable for health satisfaction accounts for overall health status. Demographic characteristics are captured by variables for the number of children, the presence of a baby in the household and for the person’s gender, religious affiliation, migration background, and age. The type of relationship is controlled for by variables for relationship duration and being married to the partner. For persons not married to the partner, we take into account whether or not the couple lives together in the same dwelling. We also include a variable for the number of previous marriages (and, hence, the number of previous divorces). Moreover, as East Germans appear to have more equal gender roles than West Germans, we control for residing in East Germany. Footnote 7 Finally, cohort dummies are included. In regressions with more than one wave of the data, we also control for the year of observation.

Leave a Reply